Understanding the dynamics of multiphase flows to enable their accurate and efficient prediction is of central importance in the oil-and-gas industry. Despite the significant advances that have been made in modelling these flows over a number of decades, however, numerous challenges, and open problems, remain. We present examples of progress made recently in the Multi-scale Examination of MultiPHase physIcs in flowS (MEMPHIS) programme aimed at providing reliable predictive tools. We also highlight the limitations of purely CFD-based approaches, and provide our perspective on potential predictive strategies based on a ‘multi-fidelity’ approach: a true fusion between model-based and data-driven modelling.
Multiphase flows and associated phenomena are ubiquitous in the oil-and-gas industry. As highlighted by the work of Shippen and Bailey  "a monumental amount of research and development work has been invested in multiphase flow modelling over the past 50 years." According to the classification of Brill and Arirachakaran , we have witnessed a transition from a purely "empirical period" from 1950–1975, which relied on the exclusive use of correlations, to the present "modelling years" era that started in the 1980s wherein multi-fluid models gained prominence; the latter combined momentum conservation equations and flow regime transitions. These models incorporated force balances, accounted for the need to satisfy certain stability and well-posedness criteria, but, nonetheless featured numerous empirical, closure relations .